NVIDIA AI Infrastructure and Operations Operations and Monitoring NCA-AIIO
Quiz 4 Question 4 of 20

An operations administrator manages a high-performance computing cluster dedicated to training LLMs. To implement a predictive maintenance strategy that flags failing hardware before it causes training jobs to crash, which GPU telemetry metric should the administrator monitor most closely as an early indicator of physical hardware degradation?

Select an answer to reveal the explanation.